How to study Machine Learning — Semester VI, Pokhara University BECE
What it covers
Algorithms that learn from data: supervised learning (regression and classification), unsupervised learning and clustering, model evaluation, neural networks, and practical machine learning workflows.
Topics worth extra time
- Supervised learning: regression and classification
- Decision trees, Naive Bayes, k-NN and support vector machines
- Model evaluation: bias and variance, cross-validation and metrics
- Unsupervised learning: clustering and dimensionality reduction
- Neural networks and the basics of deep learning
How to study it
Implement simple algorithms from scratch once (linear regression, k-NN) to understand them, then use scikit-learn on real datasets. Calculate small examples by hand, such as building a decision tree with information gain.
In the exam
Numerical questions (information gain, Naive Bayes probabilities, distance calculations) need clear tables. Explain each algorithm with its advantages, limitations and a use case.